The starting point is one of the oldest results in celestial mechanics. In 1772, Joseph-Louis Lagrange demonstrated that in the three-body problem, there exists a configuration in which three bodies, placed at the vertices of an equilateral triangle, rotate together as a rigid structure. This relative equilibrium gives rise to the famous Lagrange points L4 and L5, the triangular equilibria that today host swarms of Trojan asteroids leading and trailing Jupiter around the Sun. For masses below a critical threshold known as Gascheau’s value, or Routh’s critical mass ratio, these equilateral configurations are linearly stable: a small nudge produces a small oscillation rather than a catastrophic departure. For the Sun–Jupiter system, the margin is comfortable, which is why Trojans have persisted since the dawn of the Solar System.

But the Lagrange solution is only the seed of a much larger structure. Associated with the vertical eigenvectors of the linearized system around the Lagrange equilibrium — the modes that describe motion out of the orbital plane — there exist spatial quasi-periodic orbits. When viewed in a frame rotating with the bodies, these quasi-periodic motions become strictly periodic. In 2009, Marchal applied an averaging procedure to eliminate the fast orbital frequencies and showed that, in this averaged restricted problem, these out-of-plane periodic orbits are fixed points that form a continuous one-parameter family. Remarkably, this family connects L4 to L5, threading a bridge of inclined configurations that begins with a Trojan-like body sixty degrees ahead of its host planet, sweeps through a quadrupole of ever higher mutual inclinations — including the remarkable case where the two bodies share the same orbit but are pitched ninety degrees out of the plane — and descends again to the mirror-image equilibrium sixty degrees behind.

What Prieur and Robutel set out to determine was whether this family is an artifact of the averaging approximation, or a genuine feature of gravitational dynamics. Their answer, obtained through a combination of analytic perturbation theory and exhaustive numerical computation, is emphatic: the family persists. In the first part of the paper, they work in the planetary limit of the averaged three-body problem, the regime in which one mass dominates the other two, as the Sun dominates the planets. Using perturbation methods, they prove the persistence of Marchal’s family for nonzero masses of the two small bodies, and they derive an analytical approximation of the family that remains valid for mutual inclinations up to about sixty degrees. The approximation involves expanding the libration frequency and the precession frequency of the orbital nodes in power series of the sine of half the mutual inclination, with coefficients expressed through incomplete elliptic integrals of the first kind. These series, carried to twentieth order, provide an accurate description of the family across the low-to-moderate inclination regime where most real co-orbital systems are expected to live.

The second part of the study is purely numerical, and this is where the result becomes truly striking. Abandoning both the restricted approximation and the averaging, the authors integrated the full three-body problem — complete, non-averaged, with all three masses finite — using a continuation method that traces the family as a parameter varies. They found that the family of inclined periodic orbits persists over a wide range of masses, extending well beyond the planetary case in which the two small bodies are negligible compared with the dominant one. Tests with mass ratios between the two smaller bodies spanning four orders of magnitude confirmed that the qualitative structure of the family is insensitive to how the small mass is partitioned; only a mild symmetry breaking between the two planets appears as their mass ratio departs from unity. In other words, Marchal’s bridge from L4 to L5 is woven into the fabric of the full gravitational problem, not merely into its blurred, time-averaged portrait.

The stability analysis along the family yields the study’s most provocative finding. As one follows the family away from the planar Lagrange equilibria, the linear stability of the orbits changes character, and the inclined configurations remain stable for parameter values where the planar Lagrange triangles themselves would be unstable. Specifically, the inclined members of the family remain linearly stable even when the masses exceed Gascheau’s critical value, the threshold beyond which the classical equilateral solution destabilizes. Inclination, it turns out, is not merely a perturbation to be tolerated — it can be a stabilizing resource. A pair of planets sharing an orbit at high mutual inclination can, in principle, be dynamically sturdier than the same pair confined to the same plane. The stability transition along the family occurs at a well-defined mutual inclination, and the authors chart precisely where, along the one-parameter continuum, linear stability is gained and lost.

The implications extend beyond linear analysis into the global architecture of the co-orbital region. Stable periodic orbits act as organizing centers of phase space: around them gather invariant tori of quasi-periodic motion, and their stable and unstable manifolds sculpt the boundaries between regions of regular and chaotic dynamics. Prieur and Robutel show that the stability of Marchal’s family shapes the global dynamics of the co-orbital region, determining how much of phase space is hospitable to long-lived inclined co-orbital motion. But they also delineate a dramatic frontier. Beyond mutual inclinations of approximately sixty degrees, the family becomes highly unstable. In that regime, high-inclination co-orbital configurations are fragile, prone to rapid destabilization — a result with echoes of the Kozai-Lidov mechanism, in which large mutual inclinations exchange inclination for eccentricity and drive orbits toward close encounters. The sixty-degree boundary thus marks a fundamental divide in the geography of co-orbital phase space: below it, inclined companions can endure; above it, they are living on borrowed time.

Technically, the work required careful handling of the symmetries and degeneracies that haunt the three-body problem. Because the problem enjoys invariances — translations, rotations, and the conservation of angular momentum — the reduced phase space on which the dynamics genuinely lives is smaller than the raw coordinates suggest. The authors employed reductions based on Jacobi coordinates and the angular momentum integral, and they discuss the subtleties of achieving local uniqueness of solutions, comparing coordinate changes against the introduction of Poincaré sections that fix one body at the origin. Numerical continuation along the family, they note, can stumble at turning points where the family parameter folds back on itself, and techniques such as pseudo-arc-length continuation exist for such obstacles — though the authors found they were not needed for the results presented. The numerical code used to generate the study’s figures has been released publicly through a Zenodo repository, allowing other researchers to reproduce and extend the analysis.

The study also bridges theory and observation in a field that is rapidly maturing. Co-orbital configurations — Trojan planets, quasi-satellites, horseshoe systems — have long been hunted in exoplanetary data, and detection strategies based on radial velocity signatures of quasi-circular co-orbital planets have been proposed. Understanding which inclinations permit stable co-orbital companions directly informs where such systems might plausibly survive and therefore where observers should look. Similarly, the inclined dynamics of Neptune’s Trojans and the resonant structure of Jupiter’s Trojan swarms have been studied extensively through numerical surveys; the new analytical approximations provide a theoretical backbone for interpreting why inclined Trojans survive at low-to-moderate inclinations but thin out dramatically beyond the sixty-degree threshold. Earlier numerical studies had hinted at instabilities for highly inclined co-orbital bodies without identifying their dynamical cause; the new work supplies the missing mechanism by connecting them to the fate of Marchal’s family.

There is a pleasing historical symmetry in the result. Marchal’s 2009 analysis built on the averaging methods that trace back through the French school of celestial mechanics — the same tradition that produced Lagrange’s equilateral solution itself, as well as Liouville’s and Gascheau’s nineteenth-century studies of its stability. The new paper, the first in what the authors signal is a series, closes a loop by demonstrating that a structure glimpsed in the averaged, restricted idealization is, in fact, a permanent fixture of the exact problem that Lagrange and his successors studied. It also connects to a broader contemporary renaissance in three-body dynamics, from the discovery of the figure-eight choreography for equal masses to recent proofs of Marchal’s conjecture linking the Lagrange triangle to that remarkable orbit.

What emerges is a revised picture of co-orbital architecture. The two Lagrange points L4 and L5 are not isolated havens but the planar endpoints of an entire inclined family of stable resonant configurations, an arch spanning the co-orbital region through three-dimensional space. Planets, asteroids, or spacecraft that share an orbit need not hug the orbital plane; they can climb the arch to substantial inclinations and still remain locked in stable mutual motion — provided they stay below the sixty-degree precipice. For mission designers contemplating co-orbital spacecraft formations, for exoplanet hunters weighing the plausibility of inclined Trojan planets, and for dynamicists reconstructing the early histories of planetary systems, the map of the co-orbital region has just gained a whole new dimension.